Reversing — AI skill for Claude Code

reversing

A big skill for **reverse engineering** — the discipline AI agents need when there is no source: CTF binaries, crackmes, unknown APKs, packed executables, custom VMs, firmware, protocols, and defensive malware analysis. Born from merging the best RE skills found in the open ecosystem (the stage-gated router packs, the CTF technique corpora, the agent methodology frameworks, the VM playbooks, and the emulation-first toolkits) into one discipline: identify before interpreting, spend cheap probes first, let the program compute the answer, and prove every claim with an address.

Most reversing work is not "read all the assembly". It is a loop: triage what
the target IS, probe the cheapest observable, form one hypothesis, test it,
and escalate only when the cheap route stalls. This skill makes both failure
modes impossible: skipping triage, and grinding a stalled approach.

O que é (PT-BR)

Uma big skill de **engenharia reversa** para agentes de IA — para quando não existe código-fonte: binários de CTF, crackmes, APKs desconhecidos, executables empacotados, VMs customizadas, firmware, protocolos e análise defensiva de malware. Síntese profissional das melhores skills de REA do ecossistema aberto (zhaoxuya520/reverse-skill, ljagiello/ctf-skills, wshobson/agents, yaklang/hack-skills, p4nda0s/reverse-skills + conceito do Trail of Bits), com licenças verificadas 1:1 e crédito em todas as superfícies — o mesmo processo das skills anteriores da coleção.

What's inside

Piece	What it does
The reversing contract (SKILL.md)	Six rules: identify before you interpret, cheap before expensive, let the program compute, answer the question asked, prove it or label it, respect the boundary
The method (references/the-method.md)	The stage-gated workflow (triage → static → dynamic → synthesis), the quick-wins ladder, the evidence ladder (E1-E4), honesty rules ("never pipe triage through head"), loop prevention, completion